Research on Communication and Perception Signal Classification Based on BP Neural Network

The article considers the waveform classification and coexistence between communication signal and radar signal in the complicated scenarios. For a given performance requirement of complex communication scene, this article designs classification and predictive models such that the detection probability of the communication and radar signal is maximized. While the original detection algorithm is not ideal, five representative features are extracted and detailed Time-Frequency analysis is done to obtain a low-error-rate and stable classification method. The proposed model based on BP neutral network is designed to be robust to original imperfect and unitary algorithm. Simulation results verify that the signals can be perfectly classified and the accuracy rate is more than 95% with this proposed approach.